Systematic review of rheumatic disease phenotypes and outcomes in the Indigenous populations of Canada, the USA, Australia and New Zealand

We performed a broad search using medical literature databases and Indigenous specific online indexes and organization websites identified with the help of a medical librarian. Medical literature databases searched included Medline (1946–June 2015), EMBASE (1980–June 2015) and CINAHL (1996–June 2015). Indigenous specific online indexes and organization websites searched (June 2015) were the Circumpolar Health Database, Health Info Net, Metis Health Database, Native Health Database, Native Indigenous Studies Portal and The First Nations Periodical Index. We also did a search of each country’s government websites for relevant publications. References of relevant identified studies were reviewed for additional primary references.

This study was part of a larger review to characterize the epidemiology, clinical outcomes and healthcare service utilization of arthritis conditions for Indigenous populations of Australia, Canada, New Zealand and the USA. Keywords and Medical Subject Headings (MESH) for the terms ‘arthritis’ and ‘indigenous populations’ were selected with the assistance of a medical librarian. An example of the search strategy (conducted in Medline) is provided in ‘Appendix.’ We used an expanded version of a ‘3E’ search strategy developed for identifying studies on inflammatory arthritis [8], including validated terms that have been used to identify other arthritis conditions such as osteoarthritis [9], gout [10], juvenile idiopathic arthritis, systemic lupus erythematosus [11], scleroderma, polymyositis, dermatomyositis, Sjögren’s syndrome, as well as terms for general arthritis and rheumatic disease. Search terms for Indigenous populations of interest used both global and local terminology. Only the arthritis search terms were used during the Indigenous online indexes and websites review. No language or publication date restrictions were imposed during the electronic search. The literature search was not limited by specific clinical outcomes terms.

We identified cohort, case–control and cross-sectional studies specifying estimates of clinical outcome measures in Indigenous populations [12] from Australia, Canada, New Zealand and the USA with arthritis conditions (based on the Public Health Agency of Canada definition [13]) of osteoarthritis, rheumatoid arthritis, juvenile idiopathic arthritis, ankylosing spondylitis, psoriatic arthritis, Reiter’s disease, other spondyloarthropathies, systemic lupus erythematosus, scleroderma, polymyositis, dermatomyositis, Sjögren’s syndrome or gout. Studies were included if they reported on one of the standard measures of outcomes, including, but not limited to those suggested by OMERACT (Outcome Measures in Rheumatology) for pharmacologic and complementary interventions. Clinical characteristics or features include: tender joint counts, swollen joint counts, patient global, inflammatory markers [erythrocyte sedimentation rate (ESR), C-reactive protein (CRP)], composite disease activity scores, Health Assessment Questionnaire (HAQ), morning stiffness duration, radiographic imaging (either as % with erosions or actual score of damage), Quality of life (any validated scale), serology results [rheumatoid factor (RF), anti-cyclic citrullinated peptide (anti-CCP), anti-nuclear antibody (ANA)], extra-articular features (nodules, sicca complex, vasculitis, interstitial lung disease, neuropathy, etc.), patient-reported outcome measures including visual analogue scales for pain or fatigue.

Two review authors (CB, KE) independently screened titles and abstracts and performed the full-text review. Disagreements were resolved by consensus. Review articles and articles with secondary data were included during the initial eligibility screening and when available, the primary studies were obtained and used for data extraction. The authors then extracted the data from included studies using standardized and pretested data extraction forms to collect information on country of study, Indigenous population and number, comparison population and number if relevant, study design, type of arthritis and outcomes.

A narrative synthesis was prepared given the heterogeneity of populations, study methods and outcomes reported.

This study was part of a larger initiative to synthesize epidemiology, clinical outcomes, mortality and health services use in Indigenous populations in the four countries of interest. A total of 5269 titles and abstracts were reviewed, of which 504 underwent full-text review for any of the three outcomes above, and 85 were included for extraction of clinical outcomes (Fig. 1). Nearly all the studies described outcomes in the North American populations (n = 38 American, n = 37 Canadian, plus n = 2 studies reporting both American and Canadian data), with only 4 studies from Australia and 4 studies from New Zealand. The majority of studies were in cohorts of patients with rheumatoid arthritis (n = 31) and systemic lupus erythematosus (n = 19), with studies in all rheumatic diseases (osteoarthritis n = 7, spondyloarthritis conditions n = 13, inflammatory arthritis not meeting criteria for a specific type n = 7, self-reported arthritis n = 2, scleroderma n = 4, Sjogren’s syndrome n = 1, gout n = 1, juvenile idiopathic arthritis n = 9, juvenile SLE n = 7, juvenile spondyloarthritis n = 3) except polymyositis and dermatomyositis identified.

Both identified studies utilized a cross-sectional design. Ferucci’s study recruited 9968 Southwest American Indian and Alaskan Native participants [14]. Those with arthritis, relative to those without arthritis, had worse SF-12 Physical Composite Scores (Alaska mean 43.9 vs 52.6; American Indian mean 42.7 vs 49.7; both p < 0.001) but not SF-12 Mental Health Composite Scores. In both locations, approximately 50% of the subjects with arthritis reported pain interfering with work, compared to 15–20% of those without arthritis (adjusted OR 3.4 and 3.1, respectively). In Lawrence’s analysis of the National Health Interview Survey (1989–1991), 22.6% of American Indian and Alaska Natives with arthritis reported activity limitations, compared to 17.6% of Whites, 24.5% of Blacks, 13.0% of Asian and Pacific Islanders [15]. Thus, in the American Indigenous populations self-reporting arthritis diagnoses, significant limitations in physical function are apparent and in excess to that seen in non-Indigenous populations.

Seven studies describe disease features and impact in Indigenous populations with osteoarthritis [16‐22]. Osteoarthritis was the forth leading cause of years lost to disability in American Indians [19]. The other studies in American Indians (1) report the frequency of a positive ANA in osteoarthritis being 63% in a cross-sectional study design, and in unknown titer [17]; (2) describe in a cross-sectional study swollen (mean 1.4, SD 2.0) and tender (mean 4.4, SD 4.1) joint counts, physical function (mean Health Assessment Questionnaire score 0.46) and pain (mean 5.6, SD 2.5) [16]; and (3) describe radiographic findings in a cohort study design using the Kellgren/Lawrence scoring system, along with individual-radiographic-feature scoring systems of qualitative and quantitative assessments [18]. A single study in Canadian Inuit described that all patients with osteoarthritis had mild impairment in function, frequently had mild–moderate disease activity, nearly universal confirmation of degenerative changes on radiographs and rare ANA positivity with no patients having a positive rheumatoid factor [20]. A descriptive study of an Australian Aborigine cohort characterized clinical and radiographic findings concluding that significant degenerative arthritis was uncommon, but with the majority of osteoarthritis affecting weight-bearing joints and the lumbar spine [22]. A longitudinal study of Maori and non-Maori populations undergoing hip and knee arthroplasty compared pre- and postoperative scores relative to osteoarthritis, including the Oxford and WOMAC scores, and SF-12 General Health (PH) and Mental Health scores between 2005 and 2009 [21]. Preoperative disease specific function was significantly worse in Maori (mean Oxford scores 10.10 vs 11.26 and WOMAC 76.24 vs 73.54). Although both Maori and non-Maori improved postoperatively, there were smaller overall improvements in Maori. SF-12 PH scores were similar between groups, but SF-12 MH scores were worse in Maori preoperatively, and again at 1- and 5-year time points. These results suggest variation in osteoarthritis manifestations across Indigenous populations, although with limited homogeneity in the clinical aspects studied and reported. The single study looking at the outcomes of surgical intervention highlights higher disease severity at time of procedure, thus perhaps contributing to the limited improvement gained through the surgery.

There were 31 studies identified [4, 17, 20, 23‐50], with nine that included a comparison of clinical features to a control population, mostly Caucasian, but also to some control populations characterized as non-Indigenous (Table 1).

Our review identified nine studies in Indigenous patients with juvenile idiopathic arthritis: six from Canadian First Nations [51‐56], one from Canadian Inuit [20] and two in Alaska Native [57, 58] populations. In Canadian First Nations compared to Caucasian children, RF+ polyarticular juvenile idiopathic arthritis was more frequent (42 vs 3%), whereas pauciarticular disease was less frequent (22 vs 57%) [55]. Age of onset was not different between populations, but First Nations with RF+ polyarticular disease had a higher frequency of ANA (93 vs 44%) [55]. Onset subtype was not different for on vs off-reserve populations [55]. In Vancouver’s Children’s Arthritis Program in the 1960s, comparison between First Nations and non-First Nations children found no significant differences in clinical characteristics (fever, rash, iritis, pericarditis, poor growth, steroids, ankylosis or effusions); however, First Nations had more frequent RF+ disease (46 vs 5%) [53]. In a study of Toronto’s Hospital for Sick Children cohort in years 1984–2002, the onset type in First Nations varied from that of European populations, characterized as oligoarticular persistent (10 vs 30% European), oligoarticular extended (10 vs 12% European), RF-polyarticular (40 vs 23% European), RF+ polyarticular (20 vs 2% European) and systemic (10 vs 13% European), with no cases of psoriatic or enthesitis-related disease in First Nations but with these in 12 and 8% in European children, respectively [52]. First Nations patients had the highest rate of ANA+ arthritis, but their risk of uveitis was not elevated.

In a study with data for years 1976–1980, First Nations children with juvenile idiopathic arthritis from Vancouver and Winnipeg in Canada were compared to Caucasian children [51]. The First Nations children more frequently had a polyarticular onset subtype (59 vs 27%) and RF+ (36 vs 9%). Numerically but not statistically significant differences between First Nations and Caucasian children with juvenile idiopathic arthritis included a later onset age (8.5 vs 5.3 years), more joint involvement (mean 16 vs 9), less frequent pauciarticular disease (29 vs 61%), less risk of uveitis (12 vs 27%), more frequent ANA+ (53 vs 29%) and more frequent HLA-B27 (31 vs 15%) [51]. In further studies, First Nations race and residence on reserve were correlated with worse physical function scores and longer active disease duration in univariate analysis, and residing on reserve was predictive of worse disability in pauciarticular onset disease in multivariate analysis [54]. In a small Canadian Inuit population, polyarticular (n = 1) and pauciarticular disease (n = 3) was described [20]. A chart review was performed to characterize juvenile idiopathic arthritis subtype in a group of Inupiat children (n = 2 cases of seronegative enthesitis-related arthritis, n = 1 systemic onset, n = 2 reactive arthritis, n = 1 ankylosing spondylitis), with one third of these children having documentation of iritis [58]. In Alaska Natives, onset subtypes were described as follows: early onset (<7 years) pauciarticular ANA+ 21%, older onset pauciarticular or seronegative enthesitis-related arthritis 37%, RF+ polyarticular 32% and reactive arthritis 11% [57]. In the Yupik population, 4% had early onset pauciarticular ANA+ subtype, whereas 8% had early onset pauciarticular ANA-disease, with 71% having older onset pauciarticular or seronegative enthesitis-related arthritis, and 17% having reactive arthritis or ankylosing spondylitis [57]. In the Inupiat population, 20% had systemic disease, 33% had older onset pauciarticular or seronegative enthesitis-related arthritis, 33% had reactive arthritis, and 20% had ankylosing spondylitis [57]. There is a clear pattern of increased frequency of the polyarticular juvenile idiopathic arthritis subtype in Indigenous children of lower latitudes compared to other population groups, and a higher frequency of autoantibody positivity, yet with no increase in risk of eye complications; in contrast Indigenous children from northern populations have a predominant phenotype of pauciarticular and enthesitis-related arthritis, similar to findings in adult populations described later in this manuscript.

Seven studies describe either disease characteristics or disease activity in groups of Indigenous patients who did not meet classification criteria for a specific rheumatic disease at the time of study [16, 17, 37, 42, 59‐61]. Canadian First Nations with inflammatory arthritis were younger and more likely to be seropositive, had higher DAS28-3ESR scores [59] and were less likely to be in remission after 12 months compared to non-First Nations (23 vs 48%, significant) [61]. In a study of descriptions of joint pain in American Indians with inflammatory joint disease (n = 12), the mean swollen joint count was 11, and the mean tender joint count was 18, with evidence of physical function impairment and high levels of pain (6.7 out of 10) described [16]. The remainder of studies described the frequency of serology findings in those with inflammatory arthritis not meeting criteria for specific rheumatic diseases. A cohort of western Canadian First Nations with episodic joint swelling were found to be frequently seropositive (35% RF+, 31% ANA+) and also has many features of connective tissue diseases [42]. A small cohort of Canadian Inuit with either a polyarticular or pauciarticular presentation were all seronegative for RF and ANA [20]. In the Oklahoma American Indian population with polyarthritis, 73% were ANA+ [17]. In the Chippewa American Indian bands with peripheral polyarthritis, 22% were RF+ and 33% were ANA+ [37]. It is interesting to consider that despite increased disease activity and the high frequency of autoantibody results, it was not possible to classify inflammatory arthritis more specifically, which may reflect patients presenting with ‘overlap’ type features and ultimately delay institution of appropriate therapy.

Thirteen studies characterizing spondyloarthropathies in Indigenous populations were identified; one of these was on ankylosing spondylitis in three First Nations populations in Canada (Bella Bella, Bella Coola and Haida) and an American Indian population (Pima) in the USA [62], one was from a Canadian Inuit population with spondyloarthritis [20], two from Canadian First Nations populations with ankylosing spondylitis (Haida) [63] and spondyloarthritis (Nuu-Chah-Nulth) [42], two were from the Navajo population with ankylosing spondylitis and reactive arthritis in the USA [64, 65], and seven included analyses on a cohort of Alaskan Native patients with conditions including ankylosing spondylitis, reactive arthritis, psoriatic arthritis and undifferentiated spondyloarthritis [38, 66‐71].

In the Navajo population, 80% of ankylosing spondylitis patients were HLA-B27 positive and 43% had knee involvement [64]. Seventy two percent with reactive arthritis had the characteristic triad of arthritis, conjunctivitis and urethritis, 88% were HLA-B27 positive, 53% had radiographic sacroiliitis, and 33% had uveitis or iritis [65]. The publication on the Nuu-Chah-Nulth population described patients with sacroiliitis and peripheral arthritis in the absence of extra-articular findings and without confirmation of ankylosing spondylitis, as well as one case of reactive arthritis with peripheral arthritis and urethritis features [42]. In 10 males with ankylosing spondylitis from the Haida population, 90% had Grade 2–3 radiographic changes by the Carter scale, 80% had a history of peripheral joint symptoms, and 30% were confirmed to have iritis, whereas another 20% had a history of eye inflammation [63]. Canadian Inuit with either spondyloarthropathies were characterized for clinical features; the majority had mild functional class and disease activity limitations [20].

In the Alaska Native population, 5-year follow-up on the original set of cases is described [69]. All ankylosing spondylitis patients had Grade 2–4 sacroiliitis on radiography compared to only 33% of undifferentiated spondyloarthritis and 57% of reactive arthritis patients. When combining all cases of spondyloarthritis, over two thirds had loss of spinal motion, 42% had limited chest expansion, and 78% had peripheral inflammatory arthritis, with the knee being most commonly involved. Iritis or uveitis affected 13% of spondyloarthritis patients (36% of those with ankylosing spondylitis). In this publication, patients were also assigned to a severity category of disease (mild, moderate or severe), with modified HAQ scores varying from 0.2 to 0.9, the Dougados Functional Index Scores ranging from 1.8 to 8.4, and physician global scores ranging from 1.8 to 4.8 across severity categories, although the timing of assessment was not specified. This longitudinal study is helpful to confirm the severe impact of spondyloarthritis conditions in the Alaska Native population, as many of the studies were descriptive in nature and without comparison populations, limiting interpretations that can be drawn from the research.

In a Canadian Inuit population, approximately half of those with juvenile-onset spondyloarthritis who had radiographs had Grade IV changes [20]. From another Canadian study, 103 patients with seronegative juvenile spondyloarthritis were identified; First Nations patients represented 9% of psoriatic arthritis, 19% of seronegative enthesitis and arthritis, 44% of ankylosing spondylitis and 67% of reactive arthritis cases, with no cases of inflammatory bowel disease-related spondyloarthritis [56]. In a study from 1976 to 1980, First Nations children with spondyloarthritis from Vancouver and Winnipeg were compared to Caucasian children [51]. Onset age and the frequency of HLA-B27 positivity were similar between populations (70 vs 64%), and joint involvement was slightly higher in First Nations children but not significantly (mean 6.6 vs 4.6); however, First Nations children had more frequent eye inflammation (36 vs 4%) [51].

A single study comparing demographic and disease features of gout in 342 New Zealand Maori and 315 Europeans was identified [72]. Maori were younger at onset (46 vs 50 years), with 90% of those with a polyarticular course being Maori, and tophi in 1.2% of Maori and 0.3% of Europeans.

There were 19 systemic lupus erythematosus studies identified [5, 17, 42, 57, 73‐87] with data for Indigenous populations in all four countries available. Age at onset of disease was similar in Indigenous and Caucasian groups in both Canada [73] and Australia [87]. In the Oklahoma American Indian population, the average number of ACR classification criteria met was 5.3 (range 4–7) [17]. In an Australian Aborigine population, the mean number of ACR criteria met was 5 (range 4–7) compared to 7 (range 5–8) in the Caucasian comparison cohort [86]. Puar did not specify the number of criteria met, but noted there was no difference between the Canadian First Nations and Caucasian populations [77]. In the 1000 Canadian Faces of Lupus study, the mean number of ACR criteria met was 5.7 (SD 1.7) in Aboriginals, and 6.0 (1.7) in Caucasians, however, not significantly different [73].

Seven studies in juvenile systemic lupus erythematosus were identified [88‐94]. In a cross-sectional study of four Canadian pediatric rheumatology centers, Aboriginal patients had longer disease duration (4.3 vs 2.3 years) than other ethnicities, despite similar mean age at study [90]. Compared to White children, Aboriginal children with juvenile systemic lupus erythematosus had a significantly lower frequency of malar rash (33 vs 78%) and more frequent serositis (44 vs 11%), with no significant differences in the frequency of autoantibodies [90]. Disease activity indices (e.g., SLEDAI-2K, SLAM-R), a damage index (SDI), physician global evaluation and fatigue scores were similar across ethnicities, and health-related quality of life was not demonstrated to be significantly different among specific ethnicity groups but with limitations of unbalanced and small sample sizes [90, 94]. In the 1000 Canadian Faces of Lupus cohort, Aboriginal children with juvenile systemic lupus erythematosus had significantly elevated odds of developing serositis (OR 18.5, 95% CI 1.8–188.6) in multivariate analysis [91]. In a retrospective review of 22 First Nations children with juvenile systemic lupus erythematosus attending a single center in Vancouver, all five had lung involvement [93]. A publication in 2006 from this same center reported significantly higher frequency of manifestations of non-erosive arthritis (100 vs 32%), myositis (33 vs 0%), gastrointestinal symptoms (93 vs 9%) and the autoantibody anti-SSA (100 vs 53%) in First Nations compared to non-First Nations children with juvenile systemic lupus erythematosus, but with disease activity at presentation and damage at 6 months not being significantly different [89]. In an analysis of Medicaid enrollees from the USA, American Indians with juvenile systemic lupus erythematosus had a higher frequency of lupus nephritis, characterized by an incident rate of 1.61 (95% CI 0.72–3.58), compared to 0.30 (95% CI 0.21–0.43) in Whites [92]. Finally, in a cohort of New Zealand Maori and European children with systemic lupus erythematosus (years 2000–2010), there was no significant difference in age at diagnosis across ethnic groups [88]. In this small sample, there were statistically significant differences in disease phenotype, with serositis affecting 50% of Maori compared to 10% of European children, lupus nephritis affecting 75% of Maori versus 40% of European children with a higher frequency of World Health Organization class 4 or 5 lesions (50 vs 40%, respectively, although with no significant differences in disease activity [88]. In juvenile systemic lupus erythematosus, differences in phenotype do seem to exist, and longitudinal outcomes will be important to examine to understand the impact of these phenotypes on damage accrual and mortality.

Three studies were identified describing scleroderma clinical features in Canadian First Nations and American Indian populations [95‐97] (Table 4). The American Indian studies involved the same population of patients of Choctaw descent; Arnett’s study reported on 17 subjects, but did not involve a comparison to another population [96], whereas Kuwana’s study included 12 of these subjects but in comparison with Caucasian patients [97]. The mean age of disease onset was 4 years younger in Canadian First Nations compared to the non-First Nations population [95]. In Choctaw Native Americans, age at disease onset was also younger compared to the Caucasian population (53 vs 42 years) but this was not statistically different [97]. There were no significant differences in the cutaneous subtype (limited vs diffuse disease) nor mean Rodnan skin scores between Canadian First Nations and Caucasian patients [95]; the majority of the Choctaw patients had diffuse disease, not different in frequency to the Caucasian population [97]. Polyarthritis was more frequent in the Canadian First Nations population [95], but not the Choctaw American Indian population [97] relative to the Caucasian populations. Canadian First Nations had a higher mean number of gastrointestinal symptoms reported, at a mean of 5.8 versus 4.1 in the Caucasian population, as well as worse gastrointestinal symptom severity (mean 2.9 vs 1.7 on 0–10 scale) [95]. Raynaud’s phenomenon severity (mean 3.9 vs 2.8 on a 0–10 scale) was also worse in the Canadian First Nations population [95]. All studies described above as well as a study by Gaddy [17] examined the frequency of autoantibodies in patients with scleroderma, with a wide range of variation in the frequency of their presence, but with no significant differences found between Indigenous populations and Caucasian comparison groups. Thus, the available literature highlights variations in scleroderma phenotype in Canadian First Nations populations compared to Caucasians, whereas phenotype was not different in the American setting.

A single study reporting on serology in patients with rheumatic diseases from the Oklahoma American Indian population included a single patient with Sjogren’s syndrome, who was positive for ANA, Anti-Ro and RF antibodies [17].